Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02368422 2001-09-25
w0 00/60733 PCT/US00/08243
NON-LINEAR DISTORTION GENERATOR FOR
BOTH SECOND AND THIRD ORDER DISTORTION
BACKGROUND OF THE INVENTION
Field of the Invention
The present invention relates generally to communication systems employing
amplification devices. More particularly, the invention pertains to a non-
linear
predistortion or postdistortion generator for coupling in-line with an optical
receiver,
optical laser transmitter or an amplifier to minimize second and third order
distortion
caused by the signal processing.
Description of the Related Art
Analog intensity modulation of a distribution feedback (DFB) laser is a
widely used technique to transmit analog signals, such as sound or video
signals and
data, on optical fibers over a long distance. Optical detector also is widely
used in
fiber optic link. The performance of DFB lasers and optical detectors are
limited by
their distortion performance. Improving second order and third order
distortion
performance can greatly improve the entire system performance and increase the
entire system dynamic range.
Amplifiers are also widely used in many types of communication applications.
Although it is preferable to keep amplifiers within their linear range of
operation, it
has been increasingly necessary to extend the operation of amplifiers into
high power
and high frequency regions of operation. Typically, the output power of an
amplifier
is limited by the non-linearity of the active devices, including bipolar
transistors and
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-2-
FETs. These non-linearities result in distortions which are impressed upon the
signal
being amplified. Reducing the non-linear distortions of an amplifier results
in
increases of the output power, the system dynamic range and the carrier-to-
noise
ratio. Accordingly, minimizing distortions and achieving linear frequency
response
is paramount to efficient amplifier operation.
Minimizing distortion is particularly important when a series of amplifiers is
cascaded over a signal transmission path, such as a series of RF amplifiers in
a
CATV transmission system. Disposed throughout a CATV transmission system are
RF amplifiers that periodically amplify the transmitted signals to counteract
cable
attenuation and attenuation caused by passive CATV components, such as signal
sputters and equalizers. The RF amplifiers are also employed to maintain the
desired
carrier-to-noise ratio. Due to the number of RF amplifiers employed in a given
CATV transmission system, each RF amplifier must provide minimum degradation
to the transmitted signal.
Many amplifiers are subject to a wide range of ambient operating
temperatures. These temperature changes may affect the operating
characteristics of
certain electronic components within the amplifier, thereby inducing
additional
distortions. A temperature range of -40°C to +85°C is not
uncommon for many
amplifier applications in a communication environment. To ensure consistent
performance over the operating bandwidth, and to minimize resulting
distortions, an
amplifier must be designed for a broad range of ambient operating
temperatures.
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-3-
The distortions created by an amplifier which are of primary concern are
second (even) and third (odd) order harmonic intermodulation and distortions.
Prior
art amplifier designs have attempted to ameliorate the effects of even order
distortions, such as composite second order (CSO) distortion, by employing
push-
pull amplifier topologies, since the maximum second order cancellation occurs
when
equal amplitude and 180 ° phase relationship is maintained over the
entire
bandwidth. This is achieved through equal gain in both push-pull halves by
matching the operating characteristics of the active devices. In some cases,
second
order correction is still needed in order to get good CSO performance. Many
prior
art designs include the use of a separate second order distortion circuit to
provide
such the correction for CSO.
However, odd-order distortion is difficult to remedy. Odd-order distortion
characteristics of an amplifier are manifest as cross modulation (X-mod) and
composite triple beat (CTB) distortions on the signal being amplified. X-mod
occurs
when the modulated contents of one channel being transmitted interferes with
and
becomes part of an adjacent or non-adjacent channel. CTB results from the
combination of three frequencies of carriers occurring in the proximity of
each
Garner since the carriers are typically equally spaced across the frequency
bandwidth.
Of the two noted distortions, CTB becomes more problematic when increasing the
number of channels on a given CATV system. While X-mod distortion also
increases in proportion to the number of channels, the possibility of CTB is
more
dramatic due to the increased number of available combinations from among the
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-4-
total number of transmitted channels. As the number of channels transmitted by
a
communication system increases, or the channels reside close together, the odd-
order
distortion becomes a limiting factor of amplifier performance.
There are three basic ways of correcting distortion created by a non-linear
device
(NLD): 1) reduce the signal power level; 2) use a feed forward technique; and
3) use
a predistortion or postdistortion technique. The first method reduces the
signal power
level such that the NLD is operating in its linear region. However, in the
case of an RF
amplifier this results in very high power consumption for low RF output power.
The second method is the feed forward technique. Using this technique, the
input signal of the main amplification circuit is sampled and compared to the
output
signal to determine the difference between the signals. From this difference,
the
distortion component is extracted. This distortion component is then amplified
by an
auxiliary amplification circuit and combined with the output of the main
amplification
circuit such that the two distortion components cancel each other. Although
this
improves the distortion characteristics of the amplifier, the power consumed
by the
auxiliary amplification circuit is comparable to that consumed by the main
amplification
circuit. This circuitry is also complex and very temperature sensitive.
The third method is the predistortion or postdistortion technique. Depending
upon whether the compensating distortion signal is generated before the non-
linear
device or after, the respective term predistortion or postdistortion is used.
In this
22-03-2001 CA 02368422 2001-09-25 US 000008243
GIC-PT 109W0 -5-
technique, a distortion signal equal in amplitude but opposite in phase to the
distortion component generated by the amplifier circuit is estimated and
generated.
This is used to cancel the distortion at the input (for predistortion) or
output (for
postdistortion) of the amplifier, thereby improving the operating
characteristics of
the amplifier.
One such distortion design, as disclosed in U.S. Patent No. 5,703,530 and
shown in Figure 1, relies upon a traditional x-attenuation network and a delay
line
for gain compensation; and a diode pair coupled with a delay line for
distortion and
phase compensation. This circuit generates a distortion that is equal in
amplitude but
opposite in phase to the distortion introduced by the amplifier. Plots of the
distortions contributed by the distortion generator and the distortions
manifest by the
amplifier are shown in Figures 2 and 3. As shown, the distortion signal
compensates
for the distortions generated by the amplifier. However, the use of delay
Iines in
such a manner is impractical since delay lines are physically Iarge, are
difficult to
adjust and the results are inconsistent across a wide frequency range.
Additionally,
both amplitude and phase information are required for correct compensation.
The
'530 patent also states that the system disclosed therein is not ideal for
certain
application, such as predistortion for CATV RF amplifiers, due to the
excessive
losses introduced by the distortion circuit.
U.S. Patent No. 5,523,716 discloses another example of a distortion
compensation design; this design being directed to satellite communication
systems.
Due to the high power range of operation ofthe satellite system disclosed in
the '716
AMENDED SHEET
22-03-2001 CA 02368422 2001-09-25 US 000008243
GIC-PT 109W0 -Sa-
patent, the received RF signal drives the diode pair and, accordingly a
biasing circuit
is not required. Due to the extremely low signal level for CATV applications,
and
due to the much lower operating frequencies, such a design would not operate
effectively in a CATV environment.
European Patent Application No. 620661 discloses a distortion generating
circuit that includes DC biasing. However, there are two (2) separate DC
voltage
bias points, which pose a challenge for synchronized adjustment for distortion
correction.
An inline predistortion design, as disclosed in U.S. Patent No. 5,798,854,
provides compensation for NLDs by applying a predistorted signal equal in
magnitude but opposite in phase to the distortion produced by the NLD.
However,
the circuitry
AMENDED SHEET
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-6-
disclosed therein is not matched to the NLD. Additionally, the '854 patent
presents a
design that is typical of the prior art in the use of a high resistance bias
for the diodes.
This will reduce the correction efficiency and increase the effects of
temperature upon
the circuit.
Prior art designs also use separate correction circuits to correct for second
and
third order distortions if both types of corrections are required. This
increases the cost
of the overall circuit design and also generates more circuit losses.
Accordingly, there exists a need for a simple distortion generator which
counteracts the distortion created by an NLD. The circuit should not introduce
additional signal delay and should operate over a wide frequency bandwidth and
wide
ambient temperature range.
SUMMARY OF THE INVENTION
The present invention is an in-line predistortion or postdistortion generator
for
coupling in-line with an NLD to produce an output signal of useful amplitude,
but with
low composite second order, composite triple beat and cross modulation
distortions. The
distortion generator comprises an instant controlled non-linear attenuator
which utilizes
the non-linear current flowing through a pair of diodes to provide the proper
amount of
signal attenuation over the entire frequency bandwidth. The distortion
generator
circuitry is always matched to the NLD, thereby ensuring a frequency response
that is
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
predictable and predefined. The distortion generator permits selective
adjustment of
the non-linear current flowing through the diodes to create a second order
distortion.
The distortion generator also includes a temperature compensation circuit to
ensure
consistent operation throughout a wide temperature range.
Accordingly, it is an object of the present invention to provide a temperature
compensated distortion generator which minimizes composite second order, cross
modulation and composite triple beat distortions manifested by an NLD such as
an RF
amplifier, a laser diode or a photodetector.
Other objects and advantages of the of the present invention will become
apparent to those skilled in the art after reading a detailed description of
the preferred
embodiment.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 is a schematic diagram of a prior art distortion generator.
Figure 2 is a combination plot of the effect of using the outputs from the
prior
art distortion generator shown in Figure 1 with an RF amplifier.
Figure 3 is a combination plot of the effect of using the outputs from the
prior
art distortion generator shown in Figure 1 with an RF amplifier.
Figure 4 is schematic diagram of a n attenuator.
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
_g_
Figure 5 is a signal diagram of the diode non-linear current caused by the
input
voltage.
Figure 6 is a schematic diagram of the preferred embodiment of the second and
third order distortion generator of the present invention.
Figure 7 is a schematic diagram of the temperature compensation circuit.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The preferred embodiment of the present invention will be described with
reference to the drawing figures where like numerals represent like elements
throughout.
Although the preferred embodiment of the present invention will be described,
for
simplicity of explanation, as being coupled with an RF amplifier, those
skilled in the art
would clearly recognize that such a distortion generator could also be
utilized to
compensate for distortion in laser transmitters, optical detectors, and other
electronic
components which operate over a wide range of frequencies. The description
herein is
not intended to be limiting, rather it is intended to be illustrative.
The present invention will be described with reference to Figure 4, whereby a
~t attenuator network 20 is shown. The network 20 comprises a selected
configuration
of resistors Zl, Rl, R2, R3, Zo, RP. The signal source is input at signal
input 30 and the
output of the attenuator network 20 is seen across the output 95. Zl is the
source of
internal impedance which should be equal to the system impedance Za, which is
seen
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-9-
across the output 95. In an embodiment of the invention for use with a CATV
system,
the impedance values Zl and Zo are equal to 75 Ohms. Three of the resistors
Rl, R2, R3
form a ~t attenuator configuration. Preferably, the values (Y) of resistors R2
and R3 are
equal, and substantially larger than the value (X) of resistor Rl. Resistor RP
is
connected in parallel with resistor Ri.
As one skilled in the art would clearly recognize, when the following
condition
is satisfied:
X = 2Zo2 Y/ (YZ - Zo2) Equation (1)
the attenuator network 20 is matched at input and output, from DC to very high
frequencies. For one example of the attenuator when X = 7.5 and Y = 1.SK, the
power
attenuation A for this attenuator network 20 is:
Equation (2)
2(YZo/ (Y + Zo) + X)Y (YZo/ (Y + Zo))
(Y + X + YZo / (Y + Zo)) (X + (YZo / (Y + Zo)))
A =
(______________________________________________________________________________
_)2
Zo + (YZo / (Y + Zo) + X)Y
Y+X+YZo/(Y+Zo)
Under the condition when Zo « Y, (as is the case when X = 7.5 and Y = 1.5K):
A= (2 Zo / (2 Zo + X))2 Equation (3)
A (dB) = 10 lg A Equation (4)
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-10-
When X = 7.5 and Y = 1.5k, A (dB) = 0.42dB. This means the attenuator network
20
has very low insertion losses and a good frequency response. When X has a
small
variation due to the parallel of Rp, shown in Figure 4, from Equation (3)
Delta X
Delta A (dB ) = - 8.68 2Zo + X Equation (5)
z
D a 1 to X _ X X R R P - X - - R PP Equation (6)
From Equation (6):
z
D elta A (dB ) - 8.68 X Equation (7)
2Zo RP
For example, If RP = 375 ohms then:
7.5 7.5 _
Delta A (dB) = 8.68 150 375 - 0-00868dB Equation (8)
Equation (8) shows that when RP (375 ohms) is in parallel with R, (7.5 ohms),
the
attenuation will be reduced by 0.00868dB. This amount of attenuation change is
needed
for non-linear compensation for an amplifier. This example also shows that
when the
value of Rp » R1, (i.e., when Rp is 50 times larger than R,), adding RP
parallel with R,
has almost no effect on the impedance match, and the voltage drop over the Rp
is mainly
determined by the value of Rl.
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-11-
However, if a linear resistor RP is used in the attenuator network 20, there
will
be no distortion signal produced. The attenuator network 20 as shown is a
linear device.
In order for a distortion circuit to operate effectively, diodes are used to
create a non-
linear resistance. Preferably, Schottky diodes are utilized. At small current,
diode
current is exponentially proportional to the voltage across over the diode.
Thus diodes
can be used as a non-linear resistance. For non-linear applications, the
amount of
attenuation can be calculated as:
Delta A (dB) _ 8.68 X X - 8.68 X Ip Equation (9)
2Zo RP 2Zo I~
Where IP is the current flow through Rp, (the non-linear resistance). I, is
the
current flow through R,. Equation 9 provides the relationship of the
attenuation change
due to the current change in Ip. This equation is accurate over a broad
frequency range.
The relationship between the delta attenuation and a change in current is
still valid when
the resistance is a non-linear resistor. Accordingly, Equation 9 provides a
good
estimation of how much non-linear current is required for predistortion or
postdistortion
purposes.
Referring to Figure 5, when the input sinusoidal voltage wave changes from V ~
to VZ to V3, the output current changes from I, to I2 to I3 respectively. The
non-linear
current used for third order correction is:
I non - linear - Ii - 2 Iz + I3 Equation (10)
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-12-
From Equation 9, the non-linear current needed is:
I non - linear
D elta f~ non - linear correction (dB ) - $ .6 $ EquatlOri (11)
2Zo I output
Only non-linear current will be useful for predistortion or postdistortion
purposes. Equation 11 can be rewritten in the form of
Inon - linear eff
D elta f~ non - linear correction (dB ) _ $ .( $ Equation (12)
Ioutput
Inon - linear
Inon - linear eff - Equation (13)
R 1 / (2Zo)
Accordingly, Inon-lineareff In Equation 12 is the effective non-linear current
going
to the output port 114 which is shown in Figure 6. Io"~,~t in Equation 12 is
the total
current that goes to the output port 114. Equation 12 also shows that it is
the non-
linear current flowing through the diodes which causes the distortion
correction. Any
method which increases the non-linear current may increase the correction
efficiency.
Equation 13 shows that only a small part of the non-linear diode current is
effectively
being used for correction.
The n attenuator network 20 has low insertion loss and the voltage drop of the
input voltage on R, (shown in Figure 4) is proportional to the input voltage.
This
voltage may be used to drive a pair of diodes to produce non-linear current
and provide
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-13-
third order correction. The non-linear current flowing in the diodes will
cause an
attenuator to provide less attenuation at larger RF amplitudes, (i.e. when the
input signal
has a higher power). This may be used to compensate for the signal compression
caused
by amplification. Because of the relatively high value of the diode's non-
linear
resistance, the match of the attenuator network is almost unchanged. This
match will
not be changed even over temperature. Additionally, frequency response over
multi-
octave frequency bands is favorable.
The mechanisms of the second order correction circuit is also clear. If the DC
bias on each of the two diodes is different, for every RF positive circle and
negative
circle, Inon-1",e~ efe will be different. Accordingly, instead of third order
correction, this
circuit will also provide second order correction.
Referring to Figure 6, the preferred embodiment of the attenuator 100 for both
second and third order predistortion and postdistortion is shown. The
attenuator 100 of
the present invention includes several additional components that modify a
traditional
~ attenuator to achieve significantly better performance over a wide frequency
and
temperature range. The attenuator 100 has an input port 101, an output port
114 and
two bias control points 116, 123. The attenuator 100 may be used in a
predistortion
configuration with an amplifier or in a postdistortion configuration. For a
predistortion
configuration, the output port 114 is connected to the input of an amplifier.
For the
postdistortion configuration as shown in Figure 6, an output signal generated
by an
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-14-
amplifier, is applied to the input port 101. The attenuator 100 includes
resistors 105,
106, 107, 108, 112; capacitors 102, 103, 104, 111, 113, 115; diodes 109, 110,
and an
inductor 117.
In most prior art applications, an inductor is used as a phase control element
to
change the correction signal phase. However, in the present invention, the
inductor 117
is used in series with the resistor 108 to make a parallel resonance circuit
with the
forward biased diode capacitor. The inductive reactance cancels the specific
capacitive
reactance of the diodes. At the resonance frequency, the capacitance of the
diodes 109,
110 will be compensated by the inductor 117 so that the impedance between
points 118
and 119 will be purely resistive and can be calculated as follows:
R impedance between 118, 119 = L. ~ ~C * R~; Equation (14)
where L is the inductance of 117 in Henrys; C is the total forward biased
capacitor in
Farads; and R is the resistance 108 in Ohms. By carefully controlling L and C,
one may
get the following:
R;mPe~a~~ebe~wee~~~8. ~~9 = R Equation (15)
This means the capacitive effect has been totally canceled and an ideal pure
resistive
load over a very wide frequency range has been achieved.
In prior art systems, the capacitance associated with the diodes has not been
considered. In predistortion applications, Shottky diodes are forward biased,
which
results in a greater capacitance. When an RF signal is input across the
diodes, the
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-15-
average capacitance increases. Even at a bias of 0 volts, the impedance
introduced by
the diodes' capacitance may not be ignored since the capacitance in parallel
with the PN
junction of the diodes will reduce the overall voltage drop on the diodes,
thus reducing
the non-linear current produced by the diodes and the overall correction
effect.
Compensating for the capacitance associated with the diodes 109,110, the
inductor 117
resonates with the capacitance of the diodes 109, 110 at higher RF
frequencies, thus
extending the overall frequency response of the circuit.
The function of the resistors 105,106,107,108,112 and the capacitors 102,103,
104, 111, 113, 115 and inductance 117 is to form a modified ~ attenuation
network in
comparison to the ~t attenuation network 20 shown in Figure 4. The capacitors
102,
103,104,111,113, and 115 are also used for DC blocking and AC coupling. From
an
AC standpoint, the parallel combination of resistors 105 and 106 is
functionally
equivalent to resistor RZ of Figure 4. Preferably, the values of resistors 105
and 106
should be chosen such that the parallel combination is equivalent to the value
of
resistance of resistor 112, (i.e. ((R,os*Rlo~)~(Rlos+R~o6)) = Rn2). Resistor
108 is
functionally equivalent to resistor Rl of Figure 4; and the in-series
combination of
resistor 112 and capacitor 111 is functionally equivalent to resistor R3 of
Figure 4. The
value of resistor 107 has no effect on RF signal attenuation.
The other function for resistors 105, 106, and 107 is to supply a major DC
bias
to the diodes 109,110. The diodes 109, 110 are first connected in series; and
the series
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-16-
combination is connected to resistor 107 in parallel. Because resistor 107 has
a low
resistance value and is in parallel with the diodes 109, 110 the voltage drop
across the
diodes 109,110 will be primarily determined by the resistance of resistor 107.
If the DC
current flow in resistor 107 is much more than the current flow in the diodes
109, 110,
the DC voltage drop across the diode 109, 110, will be very stable and will be
insensitive to the presence or absence of a signal at the input port 101.
Three resistances 120, 121 and 122 act as a voltage divider to provide another
DC bias across the diodes 109, 110. As shown, resistor 121 is a variable
resistor to
provide a DC input bias at point 123. In this manner, the DC bias on the two
diodes
109, 110 is made unequal so that non-linear current produced by the positive
and
negative cycles of the diodes 109,110 is different. This unequal current
creates second
order distortion correction current. The correction polarity for the second
order
distortion depends upon the DC offset voltage at node 123. In this manner,
both second
and third order distortion correction may be provided.
The integrated functions of signal attenuation and diode bias supply avoid any
parasitic effects due to the introduction of additional bias circuitry. This
permits a high
frequency response and a favorable impedance match.
From a DC perspective, resistor 107, in parallel with capacitors 103 and 104,
provides a dissipative circuit to the capacitors 103, 104. Therefore, resistor
107 will
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-17-
discharge the accumulated electric charge of connected capacitors 103,104 in
every AC
cycle.
Diode 109 is connected to resistor 108 through capacitor 104 while diode 110
is
connected to resistor 108 through capacitor 103. Diode 109 is responsible for
the RF
distortion correction during the negative portion of the AC cycle, while the
diode 110
has the same function during the positive half of the AC cycle. The non-linear
current
of diode 109 charges capacitor 104, and the non-linear current of diode 110
charges
capacitor 103. Due to the configuration of the circuit, the voltage produced
on
capacitors 103 and 104 have the same value but different signs. The small
resistance
from resistor 107 connected to the capacitors 103, 104 discharges the
accumulated
electric charge during every AC cycle. As a result, there is no additional DC
voltage
drop across the capacitors 103,104 due to the input RF signals. This permits
the diode
109, 110 to provide the largest non-linear current for the correction purpose.
The present invention has several unique advantages over the prior art. This
circuit provides both second and third order correction at the same time. This
makes the
correction circuit very simple and effective. The attenuator 100 uses two low
series
resistances 107,108. Resistor 107 significantly improves the correction
efficiency and
resistor 108 provides for distortion correction with low insertion losses. Due
to the
attenuator 100 design, the voltage drop across resistor 108 fully loads the
diodes 109,
110 even under non-linear operation of the diodes 109,110. As a result,
maximum non-
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-18-
linear current is utilized for correction purposes. The present attenuator
design uses low
series resistance 108 in series with the inductor 117 to compensate for the
capacitance
of the diodes 109,110. Thus, this circuit may work over a wide frequency
range. This
correction circuit design is flexible and may be adjusted to different kinds
of RF hybrids
with different distortion characteristics. Additionally, the circuit is always
matched to
its input side and output side over wide frequency range. Finally, proper
phasing of the
distortion signals is inherent in the design, thereby avoiding additional
phase circuitry
and delay lines. This permits a circuit design which is much less complex and,
therefore, is compact and robust. Table 1 provides a listing of the components
shown in Figure 6. However, one skilled in the art would clearly recognize
that the
values shown in Table 1 are only for explanatory purposes, and should not be
considered to be limiting to the invention. For example, the value of resistor
108 may
range from approximately 2SZ to 30SZ. Likewise, the value of resistor 107 may
range
from approximately 10052 to 300052.
TABLE 1
COMPONENT ~ VALUE OR
IDENTIFICATION
102 I 0.1~ f
103 ~ 0.1~ f
104 ~ 0.1~ f
~ 105 ~ 6KS2
106 ~ 6KSZ
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-19-
COMPONENT VALUE OR
IDENTIFICATION
107 330SZ
108 7.5SZ
109 HP HSMS-2822#L30
110 HP HSMS-2822#L30
111 O.lp f
112 3KSZ
113 O.lp f
114 7552
115 O.lp f
117 1.SnH
120 2KS2
121 50052
122 2KS2
As previously described, the attenuator 100 uses the non-linear current
produced
by the diodes 109, 110 to compensate for the second order and third order
distortion
caused by an NLD. As shown, the attenuator 100 comprises capacitance,
resistance and
two diodes. The diodes are the only components that are sensitive to
temperature
change and the only components that require correction during operation over a
wide
temperature range. There are three factors which must be taken into
consideration when
operating the attenuator 100 over a wide temperature range:
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-20-
1 ) The diode operating current will change if the bias voltage remains
constant while the ambient temperature changes. Under the same input voltage
swing
at the input port 101 and the same bias voltage, more non-linear diode current
will be
created as the ambient temperature rises.
2) When the ambient temperature rises, the diode will produce less non-linear
correction current for the same input signal voltage and the same diode bias
current.
3) NLDs typically exhibit more distortion as the ambient temperature rises.
Accordingly, a higher diode non-linear current is required for correction of
the greater
distortion.
All of the temperature effects experienced by the attenuator 100 are related
to the
bias voltage. Some of the effects are additive while others are subtractive.
However,
the result is that for a given temperature, there will be an optimum bias
voltage to
produce the proper correction output. Proper temperature correction will be
achieved
when there is a predefined change of bias voltage verses temperature.
Refernng to Figure 7, the preferred embodiment of the temperature
compensation circuit 200 is shown. The temperature compensation circuit 200
controls
the bias of the diodes 109, 110 (shown in Figure 6) for optimum compensation
of the
distortion. As shown, the temperature compensation circuit 200 comprises two
transistors 206, 213; a capacitor 216; nine resistors 201, 202, 203, 204, 207,
209, 210,
214, 215; two diodes 205, 208; and a negative temperature coefficient
thermistor 211.
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-21-
The negative temperature coefficient thermistor 211 is coupled in parallel
with
resistor 210 to form a temperature linearized resistance, which is correlated
to a change
in temperature. The PNP transistor 206 provides a constant current source
through its
collector to the linearized resistor combination 210, 211. The constant
current provided
by the PNP transistor 206 induces a linearized voltage change across the
resistor
combination 210, 211 as the temperature changes. By adjusting the value of the
variable
resistor 202, the amount of constant current through the PNP transistor 206
can be
changed. Therefore, the voltage swing over temperature can be changed. The
constant
current also passes through the variable resistor 209, thereby creating a
constant voltage
drop that is used as a starting bias point for bias voltage adjustment. By
selectively
adjusting the resistance of resistors 202 and 209, any combination of voltage
swing and
starting bias voltage can be obtained. A NPN transistor 213, which is an
emitter
follower transistor, provides the control bias voltage from line 217 through
line 116 to
the attenuator 100, as shown in Figure 7. The two diodes 205 and 208 are used
to
compensate for the junction voltage of the two transistors 206, 213 which
change over
temperature.
Table 2 provides a listing of the components shown in Figure 7. However, one
skilled in the art would clearly recognize that the values shown in Table 2
are only for
example, and should not be considered to be limiting to the invention.
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-22-
TABLE 2
COMPONENT VALUE OR
IDENTIFICATION
201 16KSZ
202 3.3KSZ
203 4.7KS2
204 S OKS2
205 1N4148
206 2N3906
207 2KS2
208 1N4148
209 l .5KS2
210 2KS2
211 DKE 402N 10
212 100 SZ
213 2N3904
214 100 S2
215 3KS2
216 50~ f
It should be recognized that the present invention provides an instant voltage
controlled non-linear attenuator design combined with a bias supply for
optimum non-
linear correction efficiency and bias temperature stability. Even if the
temperature
compensation circuit 200 as disclosed herein is not utilized, the preferred
embodiment
of the present invention provides adequate distortion correction over a broad
CA 02368422 2001-09-25
WO 00/60733 PCT/US00/08243
-23-
temperature range. When the temperature compensation circuit 200 is utilized,
the
distortion compensation results can be further improved. Accordingly, a trade
off
between the performance of the compensating circuit and the complexity of the
circuit
must be weighted.
* * *
